Install
$ agentstack add skill-theseus-run-theseus-tool-authoring ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Tool Authoring
Use this skill for Tool primitive work and concrete tools.
Read First
packages/theseus-core/src/Tool.tspackages/theseus-core/src/tool/index.tspackages/theseus-core/src/tool/run.ts- For concrete tools: the relevant file in
packages/theseus-tools/src/
Contract
A tool is typed world access for the model. It declares input, output, known failure, execution, presentation, and interaction policy.
Preserve this pipeline:
- Decode raw model args with
tool.input. - Run
tool.execute, applyingtool.retryif present. - Validate success output with
tool.output. - Validate known failure with
tool.failure. - Present success or known failure as
Presentation. - Convert defects into
ToolDefect.
Known failures are tool-result content for the model, not runtime exceptions. Runtime errors are decode, output-shape, failure-shape, and defect errors.
Authoring Rules
- Use
Tool.defineTool; do not create ad hoc tool objects unless testing a boundary directly. - Use Effect Schema for
input,output, andfailure. - Use
Tool.Defaults.NoFailureonly when the tool has no declared domain failure. - Keep tool names stable and LLM-callable.
- Keep descriptions operational: what the tool does, not implementation trivia.
- Set
policy.interactionintentionally. Do not default write-like behavior to observe/read semantics. - If a tool touches the filesystem, shell, network, or process state, model that risk in policy and failure shape.
- Prefer small concrete tools over a generic "do anything" tool.
Verification
- Run the package-local test when changing
packages/theseus-tools:bun test src/from that package or rootbun run testwhen impact crosses packages. - Run root
bun run typecheckwhen signatures, schemas, exports, or Effect environments change.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: theseus-run
- Source: theseus-run/theseus
- License: MIT
- Homepage: https://www.theseus.run
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.